ideas as lifelong learning, communicative learning (Leeu-wis, 2004) and collective learning (societal learning) as well as participatory approaches have led to the development of innovation systems and processes within AKST. Inclusion of multiple knowledge bases, feedback loops and learning processes now aim to enable those involved to respond to emerging unpredictable circumstances. The concept is still evolving and requires more analysis of the agents involved, their behavior, the diverse interactions that characterize it (Spielman, 2005) as well as techniques and procedures to include actors to create knowledge for use and diffusion.
     Many analysts conclude from the experiences with in­ternational AKST that the constraints faced by agricultural organizations and systems are often institutional in nature (Byerlee and Alex, 1998) and that formal and informal or­ganizations need to closely interact. Consequently, science for agricultural development has become more inclusive, consultative and participatory. It reveals new opportunities but also new challenges, such as of responding to and engag­ing with a widening range of interest groups, agendas, pri­orities and opportunities. According to the CGIAR Science Council (2005) (in accordance with OECD, 2005abc) "such a systems perspective on agricultural innovation offers the potential of realizing the promise of science and technology in the context of socio-economic development and merits increased investment in future."

4.4.2 Alternatives in integration
There are two dominant types of disciplinary integration, both appearing increasingly within agricultural sciences. The first is integration of two or more disciplinary tradi­tions to form a new discipline involving formulation of new theoretical grounds and methodologies. Ecological econom­ics is one example. The second type is constructive interac­tion among separate disciplines.
     Historical   evidence   suggests   that   interdisciplinary communication and interaction often plays a key role in

the emergence of new research fields, i.e., in scientific re­newal and development. Thinking collectively about com­plex problems requires crossing boundaries both horizontally (across disciplines) and vertically (involving policy-makers, experts, practitioners, public) (Klein, 2004). This leads to par­ticipatory approaches and transdisciplinarity and thus problem solving that crosses both disciplinary boundaries and sectors of society (Scholz and Marks, 2001). It can also involve efforts towards a new unifying theory. For example, it has been proposed that agroecology could be developed and defined as an embracing discipline for studies on the entire agrifood system in all its ecological, economic and social dimensions (e.g., Dalgaard et al., 2003; Francis et al., 2003).
     Constructive interaction among disciplines does not, however, necessarily imply a genesis of a new discipline. In fact, the continuous emergence of new disciplines would merely result in the continuous reconstruction of new bound­aries to be overcome.8 The greatest value of any emergent, integrating discipline would be in establishing a common language and concepts for the participating researchers. On the other hand, interdisciplinary studies benefit from the ac­cumulated knowledge, methodologies and traditions of the contributing disciplines. In many cases an interdisciplinary orientation would supply a broader and more flexible selec­tion of the expertise and methods required for a sound re­sult than would reliance on the creation of new disciplinary approaches (Heemskerk et al., 2003; Lele and Norgaard, 2005; Kahiluoto et al., 2006). The short time frame of one study and the continuously evolving research needs and ob­jectives underline this conclusion.

8 This would be the case even if the development of the new disciplines would be based on the unifying and expanding "rhizome model" rather than the more commonly used hierar­chical model, which involves branching into distinct, semiau-tonomous fields of enquiry (Bruun et al., 2005).